With the development of the network, in the 3rd Generation Partnership Project (3GPP), manufacturers are researching the Long Term Evolved (LTE)/System Architecture Evolved (SAE) actively. As shown in FIG. 1, the LTE/SAE architecture includes: (1) a Mobility Management Entity (MME) 11, configured to store mobility management context of a UE, for example, user identifier, mobility management state and location information, to handle Non Access Stratum (NAS) signaling, and to ensure security of the NAS signaling; and (2) an SAE gateway (GW), including a Serving Gateway (S-GW) 121 and a Packet Data Network (PDN) GW 122, where the S-GW and the P-GW are two logical entities which may exist on the same physical entity or different physical entities.
The S-GW stores the user-plane context of the UE, for example, an IP address and route information of the UE, and performs legal monitoring and packet data routing. The interface S11 between the S-GW and the MME is responsible for communication between the MME and the S-GW, and exchanging mobility management information and session control information of a UE.
The MME 11 works together with an Evolved-Universal Terrestrial Radio Access Network (E-UTRAN) to implement the control-plane connection through the S1-MME interface. The S-GW works together with the E-UTRAN to implement the user-plane connection through the S1-U interface. The MME 11 is connected with the 2G/3G Serving GPRS Supporting Node (SGSN) through an S3 interface, and serves as a mobility control-plane anchor between the 3G network and the SAE network of the UE. The S-GW is connected with the 2G/3G SGSN through an S4 interface, and serves as a mobility user-plane anchor between the 3G network and the SAE network of the UE.
The P-GW 122 serves as a user-plane anchor for a UE to access PDN, communicates with an external PDN through an SGi reference point, and implements packet routing and forwarding, policy and charging enhancement, and packet filtering based on each user. The P-GW 122 is connected with the S-GW 121 through an S5 or S8 interface (in the case of roaming) to transmit the bearer control information such as bearer creation, bearer modification and bearer deletion, and to route the packet data.
A Policy and Charging Rules Function (PCRF) 13 transmits Quality of Service (QoS) and charging policy control information to the P-GW through an S7 interface.
A concept of Temporary Mobile Subscriber Identifier (TMSI) is involved both in SAE network and Universal Mobile Telecommunications System (UMTS). In a Circuit Switched (CS) domain, the identifier is known as TMSI; in a Packet Switched (PS) domain, the identifier is known as Packet TMSI (P-TMSI). The TMSI is designed to prevent a user from being tracked when the user's International Mobile Subscriber Identifier (IMSI) is exposed at an air interface, which may result in infringement on the user privacy. Therefore, after the user is attached to the network, the SGSN or the Mobile Switching Center (MSC) allocates a TMSI or a P-TMSI to a UE. For example, the SGSN allocates a P-TMSI to the UE, and the MSC allocates a TMSI to the UE. The TMSI is unique in a Location Area (LA) of the UE, or the P-TMSI is unique in a Routing Area (RA), where one LA may have several RAs. When the user accesses the network, a TMSI or a P-TMSI may be used as the identifier of the UE. When the user performs downlink paging, the user may be paged by a TMSI or a P-TMSI. If the UE discovers the paging information with the UE's TMSI or P-TMSI on the paging channel, the UE initiates the access.
When the UE accesses a new core network node, if no Iu-flex concept is introduced, the new node searches for the old node to obtain the context of the UE according to an LA Identifier (LAI) or an RA identifier (RAI). Due to existence of the Iu-flex, the LAI and the RAI are not enough for searching the old node. Thus, function of the TMSI or P-TMSI is further adopted together with LAI/RAI to determine the old node. The Iu-flex means that a many-to-many relation exists between access network device(s) and core network device(s) on the Iu interface. For example, an RNC is connected with many SGSNs, and an SGSN may access many RNCs. Multiple SGSNs constitute a resource pool. In a resource pool, multiple core network nodes (such as SGSN) are connected with all Radio Access Network (RAN) nodes (such as RNC) in the resource pool. In the traditional mode, however, one access network node is connected with only one core network node.
Furthermore, in the case of Iu-flex, the RAN may find the node, which the UE registers with, according to the information in the TMSI/P-TMSI, as detailed below.
The TMSI/P-TMSI includes 0 to 10 configurable bits which may serve as a Network Resource Identifier (NRI). The NRI is used to distinguish different core network nodes in a resource pool. When the UE accesses the pool for the first time, the RAN node is unable to find the corresponding NAS node through the NAS Node Selection Function (NNSF), so the RAN node selects a proper core network node according to principle(s) such as load sharing. After registering with a CN node in the resource pool, when moving in the resource pool, the UE does not change the CN node. The principle is as follows: The core network node, which the UE registers to, allocates a TMSI or a P-TMSI to the UE, and the TMSI or P-TMSI carries an NRI that represents the core network node. In this way, when the UE attempts to access, the UE sends an Initial Direct Transfer (DT) message to the RAN, where the message carries the TMSI or P-TMSI. The RAN node selects the previously registered core network node corresponding to the NRI in the received TMSI or P-TMSI. Therefore, the UE moves within the resource pool, while the core network node keeps unchanged. Nevertheless, when the UE moves out of the resource pool, the RAN node is unable to find the CN node with corresponding NRI, the RAN node reselects a new core network node, and the UE moves within the new resource pool, still with the core network node keeping unchanged.
The prior TMSI or P-TMSI is consisted of 32 bits, including: several (generally two) bits for distinguishing PS domain and CS domain, configurable 0-10 bits for NRI (0 bit indicates no flex), several bits for a restart identifier, and several other bits. The bits may be allocated adaptively according to the network deployment.
For example, in a TMSI or P-TMSI, two bits are used to distinguish the TMSI and the P-TMSI, five bits are used as a restart identifier which prevents allocating of an allocated TMSI caused by restart of the node, seven bits are used as an NRI, and the remaining 18 bits are available for allocating a UE identifier to each core network node.
In the conventional art, a TMSI or P-TMSI is designed in a resource pools. As shown in FIG. 2, the resource pools include Pool 21, Pool 22, Pool 23, Pool 24, Pool 25, and Pool 26. The NRIs of Pool 21 are 16-20; the NRIs of Pool 22 are 11-15; the NRI of Pool 23 is 1; the NRIs of Pool 24 are 6-10; the NRIs of Pool 25 are 1-5; and the NRI of Pool 26 is 11. As shown in FIG. 2, it is assumed that Pool 21, Pool 22, Pool 24, and Pool 25 are partially overlapped; each resource pool includes five core network nodes, where the core network nodes are distinguished with different NRIs; NRI may be reused in non-adjacent pools because the NAS node selection function or the uniqueness of the TMSI of UE in the paging area is not affected. It is assumed that a maximum of one million users can be attached to each core network node, there are 12 million users in the overlapped area of the pool, and fewer users exist in other areas.
In this network, 20 core network nodes are enough for attaching 12 million users. The NRI may have five bits (because 25=32, so the NRI is available for identifying 32 core network nodes). The identifiers allocated independently by each node is consisted of 21 bits (because 1,000,000=220, so the NRI is available for identifying two million users), two bits are used for distinguishing PS domain and CS domain, and the remaining 4 (32−5−21−2) bits are used for restarting.
The SAE network still involves design of Flex. Like the prior method, in a resource pool, more than one CN node (such as MME) is connected with all RAN nodes (such as eNodeB, i.e. ENB) in the resource pool. When a UE enters a resource pool initially, a RAN node selects a CN node according to load sharing principle. In this way, the UE is always anchored at the selected CN node when moving in the pool or accessing. Both MME and S-GW can be connected with ENB in the SAE network, thus there are two concepts: MME pool and S-GW pool. Pool overlapping is also allowed in the SAE network. In the SAE network, the MME pool or the S-GW pool includes a complete Track Area (TA). TA is similar to LA or RA in UMTS network.
FIG. 3 shows allocation of TMSI in an overlapped MME pool. In FIG. 3, it is assumed that the UE allocates one TA at a time. When the UE accesses MME pool 1 for the first time (for example, the UE enters ENB 1), an MME is selected from the MME Pool 1 (briefly known as MP 1). While the UE moves from ENB1 to ENB 2 or ENB3, it is not necessary to change the MME. When the UE moves to ENB 4, because of no interface between ENB 4 and the MME in the source MP 1 (ENB 4 belongs to MP 2 only), it is necessary to reselect the MME in the MP 2. In FIG. 3, ENB 2 and ENB 3 belong to two MME pools. That is, ENB2 is connected with each MME of the two pools through an interface, and ENB 3 is connected with each MME of the two pools through an interface. Therefore, ENB 2 and ENB 3 are the overlap part between MME pool 1 and MME pool 2. The benefits of the overlapping are: When the UE returns from ENB 4 to ENB 3, because ENB 3 is connected with MME pool 2, it is not necessary to reselect the MME. Reselection of the MME is not required until the UE accesses ENB 1. That is, the overlapping avoids the ping-pong effect (i.e. ping-pong relocation of the MME). Provided that there is no interface between ENB 3 and MME pool 2, when the UE moves between ENB 3 and ENB 4 back and forth, a ping-pong effect is initiated.
For the TA concept, the SAE network allows allocating more than one TA to the UE, which is different from a UTMS network, in a UTMS network, only one LA or RA can be allocated to a UE. In this way, if the UE in the figure above registers with the pool and the allocated TA list includes TA 1 and TA 2, no update needs to be initiated when the UE moves between ENB 1 and ENB 2 back and forth. That is, no update needs to be initiated when the UE moves within the allocated TAs.